Artificial intelligence can improve accessibility and ensure that students with disabilities have access to rich learning opportunities.
This article was excerpted and adapted from an EDUCAUSE Exchange podcast (Gerry Bayne, producer): Judy Brewer, Carly Gerard, and Mark Hakkinen, "The Impact of AI on Accessibility," EDUCAUSE Review, November 4, 2020.
In 2015 and 2016, nearly 20 percent of undergraduate students in the United States reported having a disability.Footnote1 The real percentage is likely higher, given that many students choose not to disclose disabilities to their institutions. Their dropout rates are substantially higher and their graduation rates are significantly lower than these rates for nondisabled students.
Students with disabilities experience educational barriers that many other students do not, and they can have both visible and invisible needs. Artificial intelligence (AI) is being explored to improve and create tools for more accessible learning environments. Here are three ways AI can help these students.
#1. Accessibility in Testing
Advanced speech synthesis technologies, which are based on machine learning models, are among the more promising applications of AI for students who rely on assistive technologies.Footnote2 The quality of synthetic speech is becoming more natural and improving rapidly. For example, Educational Testing Service (ETS) used technologies from Amazon to replace some human recorded audio with synthesized speech for some supplemental test content. ETS improved the user experience for students with disabilities by reducing the turnaround time for producing alternate format materials and providing a more natural and clear text-to-speech voice for these students.
#2. Content Descriptions
Turnaround time is significant when producing things like text descriptions or a complex set of test questions for students who are legally blind or have low vision. AI techniques could be used to automatically describe images. AI-based systems could also be used to do a "first pass" at describing content. Subject matter experts could then refine the content or, depending on the quality of the description, determine whether the content should be written from scratch.
#3. Webpage Interactions
AI-based tools can also be used to help with interactions by people who are unable to see content. Tools like Apple Siri and Amazon Echo and Alexa provide ways of interacting with content through a spoken dialogue model. But there are many ways for AI features to expand. A "seeing" AI, for example, could help students who find the contents of a webpage to be too visually stimulating. Students could ask the virtual assistant to read aloud the headings on a page, allowing them to get a sense for how the page is structured, figure out where to go on the page, or skip content that is not relevant. Building this type of accessibility into the system that everyone uses—so that it simply comes onboard with every smart device—may also reduce the stigma (and possibly the cost) associated with having to purchase separate accessibility tools or apps.
Challenges and Possibilities
AI-based design and development is often driven by the needs and behaviors of the "average user," and from a user experience design perspective, people with disabilities typically fall outside of the usual experience. Automatic speech recognition (ASR) systems, for example, typically are optimized around common speech patterns, not around the speech patterns of people with speech disabilities. As a result, students who rely on ASR systems are more likely to be disadvantaged in educational and work settings where the ASR may not be optimized for them.
On the other hand, AI also holds great promise for people with disabilities. In the future, ASR systems may provide error-free closed-captioning rather than approximations. AI may also allow people with disabilities to fully control their environments—not only at home but also in the classroom and the workplace.
Full-scale automation may not yet be practical, but progress is being made. Some organizations are already using AI to assess conformance to accessibility guidelines. As this use becomes more widespread, conformance assessment will become more scalable. And as this use continues, we will find many other ways in which AI can be used to improve accessibility and ensure that students with disabilities have access to rich learning opportunities.
- National Center for Education Statistics, Institute of Education Sciences, "Postsecondary Education," chapter 3 in Digest of Education Statistics: 2019 (Washington, DC: NCES, IES, US Department of Education, 2021). Jump back to footnote 1 in the text.
- For more information about the technology needs of students with disabilities, see Dana C. Gierdowski and Joseph D. Galanek, "ECAR Study of the Technology Needs of Students with Disabilities, 2020," EDUCAUSE Review, June 1, 2020. Jump back to footnote 2 in the text.
© 2022 EDUCAUSE. The text of this work is licensed under a Creative Commons BY-NC-ND 4.0 International License.